FEAT: support glm4-0414

xinference/model/llm/llm_family.json

@@ -10946,5 +10946,48 @@
       "<|im_start|>",
       "<|im_end|>"
     ]
+  },
+  {
+    "version": 1,
+    "context_length": 32768,
+    "model_name": "glm4-0414",
+    "model_lang": [
+      "en",
+      "zh"
+    ],
+    "model_ability": [
+      "chat",
+      "tools"
+    ],
+    "model_description": "The GLM family welcomes new members, the GLM-4-32B-0414 series models, featuring 32 billion parameters. Its performance is comparable to OpenAI’s GPT series and DeepSeek’s V3/R1 series",
+    "model_specs": [
+      {
+        "model_format": "pytorch",
+        "model_size_in_billions": 9,
+        "quantizations": [
+          "none"
+        ],
+        "model_id": "THUDM/GLM-4-9B-0414"
+      },
+      {
+        "model_format": "pytorch",
+        "model_size_in_billions": 32,
+        "quantizations": [
+          "none"
+        ],
+        "model_id": "THUDM/GLM-4-32B-0414"
+      }
+    ],
+    "chat_template": "[gMASK]<sop>{%- if tools -%}<|system|>\n# 可用工具\n{% for tool in tools %}{%- set function = tool.function if tool.get(\"function\") else tool %}\n\n## {{ function.name }}\n\n{{ function | tojson(indent=4, ensure_ascii=False) }}\n在调用上述函数时，请使用 Json 格式表示调用的参数。{%- endfor %}{%- endif -%}{%- for msg in messages %}{%- if msg.role == 'system' %}<|system|>\n{{ msg.content }}{%- endif %}{%- endfor %}{%- for message in messages if message.role != 'system' %}{%- set role = message['role'] %}{%- set content = message['content'] %}{%- set meta = message.get(\"metadata\", \"\") %}{%- if role == 'user' %}<|user|>\n{{ content }}{%- elif role == 'assistant' and not meta %}<|assistant|>\n{{ content }}{%- elif role == 'assistant' and meta %}<|assistant|>{{ meta }} \n{{ content }}{%- elif role == 'observation' %}<|observation|>\n{{ content }}{%- endif %}{%- endfor %}{% if add_generation_prompt %}<|assistant|>{% endif %}",
+    "stop_token_ids": [
+      151329,
+      151336,
+      151338
+    ],
+    "stop": [
+      "<|endoftext|>",
+      "<|user|>",
+      "<|observation|>"
+    ]
   }
 ]

xinference/model/llm/llm_family_modelscope.json

@@ -8719,5 +8719,50 @@
       "<|im_start|>",
       "<|im_end|>"
     ]
+  },
+  {
+    "version": 1,
+    "context_length": 32768,
+    "model_name": "glm4-0414",
+    "model_lang": [
+      "en",
+      "zh"
+    ],
+    "model_ability": [
+      "chat",
+      "tools"
+    ],
+    "model_description": "The GLM family welcomes new members, the GLM-4-32B-0414 series models, featuring 32 billion parameters. Its performance is comparable to OpenAI’s GPT series and DeepSeek’s V3/R1 series",
+    "model_specs": [
+      {
+        "model_format": "pytorch",
+        "model_size_in_billions": 9,
+        "quantizations": [
+          "none"
+        ],
+        "model_id": "ZhipuAI/GLM-4-9B-0414",
+        "model_hub": "modelscope"
+      },
+      {
+        "model_format": "pytorch",
+        "model_size_in_billions": 32,
+        "quantizations": [
+          "none"
+        ],
+        "model_id": "ZhipuAI/GLM-4-32B-0414",
+        "model_hub": "modelscope"
+      }
+    ],
+    "chat_template": "[gMASK]<sop>{%- if tools -%}<|system|>\n# 可用工具\n{% for tool in tools %}{%- set function = tool.function if tool.get(\"function\") else tool %}\n\n## {{ function.name }}\n\n{{ function | tojson(indent=4, ensure_ascii=False) }}\n在调用上述函数时，请使用 Json 格式表示调用的参数。{%- endfor %}{%- endif -%}{%- for msg in messages %}{%- if msg.role == 'system' %}<|system|>\n{{ msg.content }}{%- endif %}{%- endfor %}{%- for message in messages if message.role != 'system' %}{%- set role = message['role'] %}{%- set content = message['content'] %}{%- set meta = message.get(\"metadata\", \"\") %}{%- if role == 'user' %}<|user|>\n{{ content }}{%- elif role == 'assistant' and not meta %}<|assistant|>\n{{ content }}{%- elif role == 'assistant' and meta %}<|assistant|>{{ meta }} \n{{ content }}{%- elif role == 'observation' %}<|observation|>\n{{ content }}{%- endif %}{%- endfor %}{% if add_generation_prompt %}<|assistant|>{% endif %}",
+    "stop_token_ids": [
+      151329,
+      151336,
+      151338
+    ],
+    "stop": [
+      "<|endoftext|>",
+      "<|user|>",
+      "<|observation|>"
+    ]
   }
 ]

xinference/model/llm/vllm/core.py

@@ -231,6 +231,9 @@ class VLLMGenerateConfig(TypedDict, total=False):
     VLLM_SUPPORTED_CHAT_MODELS.append("gemma-3-1b-it")
     VLLM_SUPPORTED_VISION_MODEL_LIST.append("gemma-3-it")
 
+if VLLM_INSTALLED and vllm.__version__ >= "0.8.4":
+    VLLM_SUPPORTED_CHAT_MODELS.append("glm4-0414")
+
 
 class VLLMModel(LLM):
     def __init__(